Environ. Sci. Technol. 2009, 43, 8542–8547
Gas/Particle Partitioning Behavior of Perfluorocarboxylic Acids with Terrestrial Aerosols H A N S P E T E R H . A R P * ,†,‡ A N D KAI-UWE GOSS§ Department of Environmental Engineering, Norwegian Geotechnical Institute (NGI), P.O. Box 3930 Ullevål Stadion, N-0806, Oslo, Norway, Institute of Biogeochemistry and Pollutant Dynamics, ETH-Zurich, Universita¨tsstrasse 16, 8092 Zurich, Switzerland, and UFZ Helmholtz Center for Environmental Research, UFZ, Permoserstr. 15, 04318 Leipzig, Germany
Received June 24, 2009. Revised manuscript received October 2, 2009. Accepted October 5, 2009.
Experimentally determined gas/particle partitioning constants, Kip, using inverse gas chromatography (IGC) are presented for perfluorocarboxylic acids (PFCAs), covering a diverse set of terrestrial aerosols over an ambient range of relative humidity (RH) and temperature. The results are compared to estimated Kip valuesusingarecentlydevelopedmodelthathasbeenvalidated for diverse neutral and ionizable organic compounds. The modeling results consistently underestimate the experimental results. This is likely due to additional partition mechanisms unique for surfactants not being accounted for in the model, namely aggregate formation and water surface adsorption. These processes likely also biased the IGC Kip measurements compared to ambient PFCA concentrations. Nevertheless, both the experimental and modeling results indicate that partitioning to terrestrial particles in ambient atmospheres is negligible, though sorption to condensed water can be substantial. This favors rain sequestration as a more important atmospheric removal mechanism than dry particle sequestration. PFCAs found on particle filters during ambient sampling are thus accountable to vapor-phase PFCAs or aqueous-phase PFCAs sorbing directly to the filters, or the trapping of perfluorocarboxylatesalt particles. Further work on understanding the partitioning and speciation of PFCAs in atmospheric water droplets is needed to further quantify and understand their atmospheric behavior. To aid in this, a general RH dependent Kip model for surfactants is presented.
Introduction Perfluorocarboxylic acids (PFCAs) are globally present in the atmosphere. They have been found ubiquitously in air (1-4) and precipitation samples (2, 5-7), and in remote arctic regions (7). Possible origins of PFCAs in the atmosphere include direct emissions of PFCA vapors and perfluorocarboxylate-salt particles from production facilities (5, 8, 9), diffuse emissions from products containing perfluorinated coatings (9, 10), transformation from volatile precursors * Corresponding author e-mail:
[email protected]; tel:++47 2202 1988. † Norwegian Geotechnical Institute (NGI). ‡ ETH-Zurich. § UFZ Helmholtz Center for Environmental Research. 8542
9
ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 43, NO. 22, 2009
(11-15), and pollutant recycling from marine aerosols on water surfaces (16). Though PFCAs are known to be in the atmosphere, how gas/particle partitioning with terrestrial aerosols influences their atmospheric transport and deposition remains uncertain. Active sampling in terrestrial and marine atmospheres consistently indicates that the amount of PFCAs found on particle filters is much higher than those found in vaporphase sorbents placed downstream from the particle filters (1-3, 8, 17-20). On this basis, it may be inferred that PFCAs readily sorb to air particles, and further that particle scavenging of PFCAs is a dominating process for their atmospheric removal (e.g., 16, 21). However, we have reported recently that at typical atmospheric concentrations PFCA vapor sorbs substantially to quartz and glass fiber filters used for particle air sampling (22, 23). It follows from this that filter extracts will include both particle and vapor-phase PFCAs, and it is impossible to differentiate between the two fractions. Thus, there is no way to quantify sorption to terrestrial aerosols using current air sampling techniques. To address this, here we present experimental and modeling studies on the partitioning behavior of PFCAs to ambient terrestrial aerosols, based on procedures recently developed for a large variety of apolar, polar, and ionizable organic compounds (23-26). It should be noted for clarity that this study specifically pertains to perfluorohexanoic acid (PFHxA, 307-24-4), perfluoroheptanoic acid (PFHeptA, 375-85-9), and perfluorooctanoic acid (PFOA, 335-67-1) and for aerosols present in typical urban and rural environments; it does not pertain to perfluorocarboxylate-salt, marine aerosols, and particles near PFCA production facilities.
Materials and Methods Chemicals. All chemicals were obtained from Fluka (Buchs, Switzerland). Measurement of Kip Values. The gas/particle partitioning coefficient, Kip, is defined as: Kip (m3 /g) ) cip /ciair
(1)
where cip is the equilibrium concentration of compound “i” sorbed to the particle phase (molip/gp) and ciair is the equilibrium concentration of i in the vapor phase (moliair/ m3air). As the weight of aerosols changes with relative humidity (RH) due to the sequestration of water, in the presented work cip and Kip are normalized to the dry aerosol mass, Mdry (gp). Kip values were measured using a recently developed technique involving inverse gas chromatography (IGC) (25), in which a flame ionization detector (FID) was used to measure the sorption of organic chemicals to ambient particles loaded on fiber filters. The fiber filters (Pallflex Fiberfilm/T60A20, Lot T7833C) were surface deactivated by silylation as described in ref 25, and therefore do not strongly sorb PFCAs as nontreated glass and quartz fiber filters do (23). Seven of the same nine aerosol samples in the previous studies (25-27) are again used in this study; these samples include urban samples (labeled Zurich, Berlin Winter, Berlin Spring), suburban samples (labeled Duebendorf Fall, Duebendorf Winter), a mineral rich sample (Duebendorf-Sahara), and a rural sample (Roost). FIDs exhibit a relatively low sensitivity for PFCAs; however, more sensitive detectors (e.g., mass analyzers, electron capture detectors) could not be used due to the high RH and flow rates required for these experiments. To obtain a resolvable FID signal it was necessary to inject relatively large quantities of PFCAs into the IGC unit, compared to the other 10.1021/es901864s CCC: $40.75
2009 American Chemical Society
Published on Web 10/15/2009
TABLE 1. PFCA and Aerosol Specific Parameters for the Dual-Phase Sorption Model (eq 2) (at 15 °C) compound specific parameters log KiWIOM (m3/g)
log Kiaw (-) PFCA
pKa
COSMOtherm
SPARC v4.2
COSMOtherm
SPARC v4.2
PFHxA PFHeptA PFOA
0-3.8 0-3.8 0-3.8
-3.41 -3.06 -2.75
-3.08 -2.56 -1.92
0.099 0.343 0.589
-0.029 0.181 0.320
aerosol specific parameters fWIOMa,a
b
aerosol sample
Mdry (mg)
COSMOtherm (-)
SPARC v4.2 (-)
pH (-)
Vw 50% RH (µL)
Vw 70% RH (µL)
Vw 90% RH (µL)
Berlin Spring Berlin Winter Zurich Roost Duebendorf-Sahara Duebendorf Fall Duebendorf Winter
4.53 ( 0.45 3.06 ( 0.31 5.19 ( 0.52 2.21 ( 0.22 5.57 ( 0.56 4.20 ( 0.42 2.43 ( 0.24
0.04 ( 0.04 0.08 ( 0.07 0.13 ( 0.09 0.08 ( 0.08 0.03 ( 0.03 0.06 ( 0.06 0.10 ( 0.09
0.08 ( 0.05 0.13 ( 0.07 0.17 ( 0.08 0.12 ( 0.09 0.05 ( 0.03 0.13 ( 0.08 0.16 ( 0.10
3 3 3 2 2 3 2
0.36 ( 0.26 0.18c 0.31 ( 0.34 0.13 ( 0.13 0.17 ( 0.37 0.13 ( 0.28 0.15c
1.59 ( 0.30
4.89 ( 0.42
0.36 ( 0.34 0.29 ( 0.14 0.33 ( 0.39 0.21 ( 0.29
3.53 ( 0.46 1.44 ( 0.18 0.84 ( 0.40 2.65 ( 0.38
a Estimated by comparing experimental Kip and estimated KiWIOM values for apolar and large polar compounds at 50% RH (see ref (24)). b Estimated from experimental Kip and Kiaw data for a diverse set of acidic and basic compounds (see ref (26)). c Not measured, assumed to be 6% of Mdry.
compounds for which this method has been validated (25-27). For these measurements, PFCA sample vials were preheated to ca. 60 °C, and headspace volumes of 2.5-5 mL of PFHxA, 10 mL of PFHeptA, and 30 mL of PFOA were injected into the IGC. Based on vapor pressure data (28), this would correspond to exposure amounts of ∼0.6 mg for each PFCA (compared to ∼25 mg of sample, including filters and aerosols). There are three possible experimental biases with this approach that need to be accounted for: (1) the occurrence of “irreversible sorption” to the aerosols, as occurred with earlier measurements for silica filter surfaces (23), in that a fraction of injected PFCA may sorb so strongly that it does not pass through the column, even after flushing with a substantial amount of carrier gas; (2) condensation of PFCA crystals in the system; (3) nonlinear partitioning processes, resulting in measured Kip values being substantially different from their value at smaller, environmentally relevant air concentrations. To investigate the first potential bias of “irreversible sorption”, replicate Kip values for a polar compound (pentanol) and an apolar compound (undecane) were determined before and after exposing the particles to PFHexA. If PFHexA was still sorbing to the particles in the system after exposure, experimental Kip values for other compounds would be expected to decrease due to surface deactivation, as they did for silica filter surfaces (23). However, after exposure to PFHexA, no significant change in the Kip values for undecane was observed, and only a minor significant increase in Kip values for pentanol was observed (by 10-40%), likely due to deliquescence of the water in the carrier gas (see the Supporting Information (SI)). Thus we conclude that PFHexA did not appear to sorb irreversibly to the terrestrial aerosols tested. This was expected based on earlier observations that water-insoluble organic matter (WIOM) and the aqueous components were the main sorption phases of terrestrial aerosols (26), which are unlikely phases for PFCAs to sorb irreversibly to (such as raw quartz and glass surfaces). The second potential bias of PFCA crystallization is considered unlikely to be occurring. The high flow rates, as well as surface area of frits and filters inside the column are not favorable for crystal formation, but more to monolayer surface coverage. Nevertheless, this potential artifact cannot be ruled out completely.
The third potential bias of nonlinear sorption is considered likely to be occurring in the IGC and biasing measurements. The most likely nonlinear partitioning process is PFCA aggregate and micelle formation in the aerosol’s aqueous components. PFOA (analogously to n-octanoic acid) is known to form premicellar aggregates at concentrations , the critical micelle concentration (cmc) (29-31). The larger the amount of PFCA present, the greater the amount of aggregation is occurring and the greater the resulting Kip. Due to the relatively large amount of PFCA required to get a resolvable FID peak, the influence of injection amount on measured Kip could not be quantified. Thus, the potential positive biases of crystallization and aggregate formation are incorporated into the discussion. Modeling of Kip Values. Kip values for most neutral and ionizable organic compounds can be modeled collectively using a dual-phase sorption equation that accounts for partitioning into the two most dominating sorption components of terrestrial aerosols, the WIOM phase and the aqueous phase (26): Kip ) fWIOMKiWIOM + VwRH /(Diaw · Mdry)
(2)
where fWIOM is the mass fraction of WIOM in dry aerosol (gWIOM/Mdry), KiWIOM is the sorption coefficient of WIOM (ciWIOM/ciair), VwRH is the volume of water in the aerosol sample (m3water) (which is dependent on the RH and the aerosol sample’s hygroscopicity) and Diaw is the dimensionless air-water distribution coefficient (ciair/ciwater): Diaw ) Kiaw · S · (1 + 10pH-pKa)-1 for organic acids (3) where Kiaw is the dimensionless air-water partition coefficient for the neutral compound (ciair/ciwater), S is a unitless empirical factor to account for deviation from the sorption behavior in pure water (S appears to be near 1 for terrestrial aerosols and thus can be ignored, (26)), and pH refers to that of the aerosol’s aqueous phase. Note this model assumes that ionizable compounds, such as PFCAs, can only ionize in the aqueous phase and not in the WIOM phase (as typically only neutral molecules exist in non polar solvents). The compound specific parameters, pKa, KiWIOM, and Kiaw for the PFCAs studied here are listed in Table 1. There remains VOL. 43, NO. 22, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
8543
some discussion on the pKa of PFCAs longer than 6 carbon units, with proposed values ranging from 0.0 to 3.8 (32-34). We strongly favor a value of 0.0 for reasons presented earlier (35), and recent experimental evidence by Cheng et al. (31), showing the pKa of monomeric PFOA is
